AI start-up Anthropic launches bug reporting scheme
Synthetic intelligence startup Anthropic launched a vulnerability disclosure program (VDP), managed by HackerOne, in August with bounty rewards as much as $15,000 for novel, common jailbreak assaults that would expose vulnerabilities in vital, high-risk domains equivalent to CBRN (chemical, organic, radiological, and nuclear) and cybersecurity.
A jailbreak assault in AI entails a technique for circumventing an AI system’s built-in security measures and moral pointers, permitting a consumer to elicit responses or behaviours from the AI system that will usually get blocked.
“As we work on growing the following technology of our AI safeguarding methods, we’re increasing our bug bounty program to introduce a brand new initiative targeted on discovering flaws within the mitigations we use to stop misuse of our fashions,” Anthropic mentioned in a weblog submit on the revamped program.
